In the last few days Bryony Gordon wrote a piece for the Telegraph about the experience of her parents getting divorced when she was 20. It was a touching piece, full of understanding about her own reaction at the time, and the realisation that all of our parents are flawed human beings. At the point I read it, there had already been hundreds of comments.

As I started reading it, I saw that number, and my heart began to sink, as I already knew what I would find “below the line”. Sure enough, it felt like around half the comments were simply using Bryony’s piece as a hook to demonise Chris Huhne and his ex-wife. The majority of the rest were telling Bryony to “get real”, get over herself, that the article wasn’t worth reading, or “my personal experience of this sort of situation was different to yours, therefore your experience was invalid”.

Charlie Brooker recently said: “I’d say that enabling reader comments is the worst thing to have happened to newspapers since … since the last worst thing that happened to newspapers. I think there’s a Letters page for a reason. There’s plenty of room on the internet for people to say what they want, and where that isn’t is tacked on the end of something somebody else has written”, and a fellow Guardian columnist Hadley Freeman seconded that on Twitter.

It sparked a very public debate by some of my ex-colleagues at the Guardian about the value of comments, and the perception of them amongst the editorial staff. It made for awkward reading for anyone who has spent a long time working on the community aspect of news websites as I have done. The Guardian recently released some interesting numbers about their community activity. In a blog post just before Christmas, readers’ editor Chris Elliot stated that “The Guardian website publishes around 600,000 comments a month, with 2,600 people posting more than 40 comments a month.” That amounts to at least 17 percent of the comments on coming from just 2,600 users.

I did some further analysis on the number in January, and it took me a matter of minutes to identify 20 commenters who between them had racked up over 6,184 comments, an average of around 300 each. In other words, one percent of that 600,000 total could be coming from just twenty people. Numbers aren’t the whole story though. I recently spoke to one person who works for the community team on a large news website, and they said the trouble is that developers often see these usage numbers as points of data. As someone who is actually in the comment threads every day, the community person said they know that, actually, this user is housebound, this user is at home caring for a disabled son, that there are life stories behind these prolific commenters, that they are more than just dots on a graph.

And I’ve got a confession to make. One of those 20 people whose comments I counted up was me. I love taking part below the line in comment threads, even when it can get a bit “robust”. I wrote once about how, when I read the printed paper, it feels frustrating that articles come to a full stop without the comments beneath. I know the debate will be continuing online, and that the written word of the journalist is no longer the final word on any article. Community interaction on websites, when it is great, can be truly great. You can crowdsource stories, discover eye-witnesses, and bring together a range of points of view from people across the whole of the globe. And you get a free gang of extra sub-editors ready to pounce on any typo. But one of the problems for general news sites is an issue of scale.

As long ago as 2007, the BBC was setting out the principle that they should “Link to discussions on the web, don’t host them: Only host web-based discussions where there is a clear rationale” This was after having spent a lot of money over the previous few years on hosting and moderating heavily used forums. Conversation doesn’t scale. The Natural History Museum has a “bug forum” where people can ask questions about bugs, and get identifications of what they are – and sometimes what to do about them – from museum scientists and enthusiastic amateur naturalists. It is a lovely little peer-to-peer niche support group.

On a general news site, however, most of the time the motivation to comment isn’t to solve a problem for someone else, but to express a view. And that view is usually that you are angry about the news that has happened – “this is a disgrace!” – or angry at the way the news is being reported – “typical of this newspaper to twist this story to fit their agenda”. In both cases, being angry about stuff isn’t conducive to conversation. And as former head of digital engagement at the Guardian Meg Pickard has often observed, if you write a provocative column, you can expect people to be provoked.

B2B publications are – or at least should be – in a happier place. If you are working in a specific trade or vertical, already your pool of potential commenters is narrowed to something more manageable, and you should mostly be dealing with experts. Nobody rushes over to Drapers Online to leave the comment “i know nothing about draping but i wouldn’t drape it like that if you paid me” in the way a certain class of commenter is drawn to point out that they didn’t go to a specific gig, and that they don’t like the band that was playing anyway, under music reviews on a more general news site.

Should media sites abandon commenting? It is clear that it doesn’t work for a good proportion of the audience. When I was user-testing the Guardian’s iPhone app as we were working towards releasing version 2, virtually all of the people we saw were horrified that we intended to add comments into the app, explaining that the reason they preferred reading the website in the app was precisely because it didn’t have the comments.

You can’t measure the brand damage toxic comments underneath articles is doing amongst the silent majority who don’t comment, nor the obvious morale impact it is having on staff, who see their employer allowing people to describe their work as rubbish under every single thing they file. The evidence is that having comments produces a more “engaged” audience. Articles with comments on generate longer page “dwell” times on average, and that gives you a better audience profile to sell to advertisers. TechCrunch and Politico may have thought they could solve their community problems with new software, but it is community management that so many media websites are lacking. Amongst all the odious comments underneath that Bryony Gordon piece, there were some lovely supportive comments, some thanking her for the article, some explaining that they went through a similar experience. It could have been so much closer to the supportive environment of that “bug forum”. A more pro-active “gardening” of comments to highlight the kind of behaviour you want to encourage might not go amiss, but of course, this takes resource.

For too many media organisations, for too long, community “strategy” has simply consisted of opening up comment forms underneath as much content as possible, and hoping for the best. It is no surprise that it doesn’t work.